Sites Inria

Version française

Computing & seismology

Cécile Michaut - 25/08/2011

Improving the forecasting of earthquakes


Although we can predict seismic movements on the surface for a given tremor, we still cannot predict the occurrence of these tremors. Very long-term research will be needed. Nathalie Glinsky, a member of the Nachos project team and of the Mediterranean Centre for Technical Equipment Studies (CETE), and Etienne Bertrand, a researcher at CETE Méditerranée, tell us more.

Every time there is an earthquake, the same question arises: why don’t we know how to predict earthquakes? Although we can predict a volcanic eruption several days in advance, the same is not true for earthquakes. "Be careful not to confuse forecasting and prediction. We can detect zones where there are major earthquakes, and understand seismic movements on the surface when we know the source and the scale of the shock: that is forecasting. On the other hand, we do not know when or where precisely the earthquake will occur, so we cannot predict."  Researchers understand the surface movements for a given earthquake, providing they know the soil type, as shockwaves spread out differently depending on the type of rock. They are trying to improve soil mechanics models through field surveys or by observing how soil samples become deformed in laboratory tests. They are also trying to better locate the sources of earthquakes and to understand the mechanics of tears on a fault line.

The probability of such a violent event was low, but not impossible.  

To evaluate the risk , namely the effects of earthquakes at a given location, seismologists set up networks which permanently record small movements in the earth (micro-seismicity). They are also making use of historical records, which sometimes go back to the Middle Ages. They need data on as long a time-scale as possible in order to improve forecasts. Nonetheless, there is still a great deal of uncertainty. "The power of the earthquake in Japan, measuring 9 on the scale, was a surprise. The probability of such an event was low, but not impossible.”  The shallow location of the earthquake’s epicentre also explains the damage.

A lot of observation will be required if earthquakes are to be better predicted. The Demeter satellite, launched in June 2004, continuously measures many physical parameters to help study earthquakes. "We are working on the instrumentation , in order to see if it is possible to correlate the disruption of certain parameters, such as radon emissions (a radioactive gas found naturally in the ground) or the earth’s electromagnetism, with the occurrence of earthquakes. At the moment, no such correlations have been found, but we have to make observations over very long periods of time."  Certain phenomena only reoccur after several centuries, so it is necessary to make observations based on several hundreds of years in order to understand them.

Keywords: Earthquakes Alerts Modelling 3-D visualisation Risk Prevention Nathalie Glinsky